The major problem in therapy of acute myeloid leukemia (AML), like most cancers, is therapeutic resistance. Although >85% of patients respond to chemotherapy, over half will relapse. Chemotherapeutic resistance implies heterogeneity among cancer cells. Two hypotheses exist to explain this heterogeneity. The stem cell hypothesis proposes a functional heterogeneity within an ordered structure in which the bulk of leukemic cells differentiate from leukemic stem cells (LSC). In this model, LSCs are intrinsically resistant to chemotherapy. Consequently, relapsed tumors arise from LSCs and their genetic diversity will reflect that of the original tumor. As such, chemotherapeutics are targeted toward the genetic lesion found within the LSC. The clonal evolution model proposes that there is heritable variation between cells, some or all of which have the potential to form a new tumor. External stressors, such as chemotherapeutics, apply evolutionary stress on the entire population, driving clonal evolution and selecting for resistant clones. In this model, (epi)genetic diversity in the original tumor will increase the probability of chemoresistance, due to the wider repertoire of mutations that could provide a selective advantage. Moreover, as tumor cells evolve and develop new mechanisms of resistance, the profile of genetic and epigenetic lesions in the relapsed tumor continues to change. Accordingly, optimal therapy will require a mix of therapeutics with diverse mechanistic bases. Here, we will test our hypothesis that chemotherapy resistance in AML is best modeled by clonal evolution and utilize our results to better understand the mechanism(s) of chemotherapy resistance in AML. In Specific Aim 1, we will use a novel xenotransplantation model of AML to determine empiricially if LSCs are enriched after chemotherapy. In Specific Aim 2, we will use innovative single cell approaches to characterize a cohort of 30 AML paired samples (de novo diagnosis vs. relapse) for clonal diversity to determine if therapy selects for clones. In Specific Aim 3, we will perform an exploratory study to determine if clonal diversity (as measured by a novel emulsion bisulfite sequencing (BBS) assay) predicts the probability of relapse in AML patients. Specific Aim 4 will identify and confirm gene mutations that are selected by chemotherapy, by performing RNA sequencing and SNP arrays to compare the level of expression, sequence of expressed genes and copy number alterations between the de novo and relapsed AML from 30 matched pairs of patient samples. Through these diverse approaches, we will test the several features of the LSC vs. clonal evolution models of chemotherapy resistance. These results will have important implications for the development of novel therapeutics for AML, the treatment of chemoresistant disease and the prevention of relapse.